www.nature.com/mi

REVIEW ARTICLE OPEN New insights into intestinal phages

R. Sausset1,2, M. A. Petit1, V. Gaboriau-Routhiau1,3,4 and M. De Paepe1

The intestinal microbiota plays important roles in human health. This last decade, the viral fraction of the intestinal microbiota, composed essentially of phages that infect bacteria, received increasing attention. Numerous novel phage families have been discovered in parallel with the development of viral metagenomics. However, since the discovery of intestinal phages by d’Hérelle in 1917, our understanding of the impact of phages on gut microbiota structure remains scarce. Changes in viral community composition have been observed in several diseases. However, whether these changes reflect a direct involvement of phages in diseases etiology or simply result from modifications in bacterial composition is currently unknown. Here we present an overview of the current knowledge in intestinal phages, their identity, lifestyles, and their possible effects on the gut microbiota. We also gather the main data on phage interactions with the immune system, with a particular emphasis on recent findings.

Mucosal Immunology (2020) 13:205–215; https://doi.org/10.1038/s41385-019-0250-5

INTRODUCTION has been variable and antibiotics proved to be both more efficient

1234567890();,: The human gut contains a large number of viruses, mostly and cost-effective, leading to the almost abandonment of phage bacteriophages, or phages, which infect bacteria. As other viruses, therapy in most countries (reviewed in ref. 8). With the rise of phages are classified according to their type of nucleic acid, capsid bacterial resistance to antibiotics, phage therapy has recently morphology - notably the presence or absence of a tail - and the regained interest, fueling researches on applied but also basic presence or not of an envelope. The genetic material of phages phage biology. The relatively recent discovery of the influence of consists of double-stranded (ds) or single-stranded (ss) DNA or phages in aquatic bacterial ecosystems further explains the RNA, and their genome sizes range from ∼3.5 kb (e.g., ssRNA present bloom of phage studies.9 Finally, due to increased genome of Escherichia coli phage MS2) to ∼540 kb (dsDNA awareness of the importance of the gut microbiota in human genome of Prevotella LAK phages). There is considerable diversity health, a growing number of studies are addressing the roles of among phages, but 95% of them are non-enveloped tailed dsDNA phages in the gut microbiota. Emerging views suggest that phages, or Caudovirales. Within this group, the traditional intestinal phages play important roles in health and disease by differentiation into Siphoviridae, Myoviridae, and Podoviridae shaping the co-occurring bacteriome, but also by interacting families, based on tail types, is not fully coherent with phylogeny, directly with the human immune system.10–12 Several recent and therefore progressively abandoned. In addition, new phage reviews have exhaustively reported different aspects of intestinal types are constantly discovered, and classification is currently phage biology, such as its genetic diversity,13,14 bacterial ongoing reorganization. resistance mechanisms, including CRISPR–Cas systems and other Phages are present in all microbial environments and the molecular mechanisms of phage–bacteria interactions,2,3 importance of phage predation on bacteria is evidenced by the phage–bacterium antagonistic interactions in the gastro- large repertoire of bacterial anti-phage defence mechanisms. Anti- intestinal tract (GIT),15,16 lysogeny,11 and phage interactions with phage systems include cell-surface modifications that prevent the host immune system.10,11,17 Here we aim at giving a global phage recognition (phage multiplication is highly dependent on view of current knowledge of phages in the GIT, emphasizing on the proper selection of their target bacteria, which is achieved by new results, open questions, and technical difficulties of this the recognition of a specific structure on the bacterial surface1), rapidly growing field of research. but also abortive infection mechanisms that trigger cell death upon phage infection and restriction-modification or CRISPR–Cas systems that cleave invading phage genomes (reviewed in refs. 2 COMPOSITION OF THE INTESTINAL PHAGEOME and 3). Description of intestinal phages, either from a taxonomic or The presence of phages in the intestine has been described lifestyle point of view, is still in its infancy compared with that of only 2 years after their discovery by Twort,4 when d’Hérelle5 intestinal bacteria, and encounters technical difficulties. First, viral independently discovered phages, and their therapeutic potential, genomes lack universal marker genes such as the 16SrRNA gene in the stools of patients with dysentery. Before the dawn of used for bacterial taxonomic assignment. Second, the genetic antibiotics, but also later on in the Soviet Union, phages have diversity of phages remains largely unknown, preventing been utilized to treat a variety of intestinal infections, mainly sequence-based identification of most intestinal phages. Typically, cholera6 and dysentery.7 However, the success of these treatments 75% to 99% of sequences from intestinal phages do not produce

1Micalis Institute, INRA, AgroParisTech, Université Paris-Saclay, 78350 Jouy-en-Josas, France; 2Myriade, 68 boulevard de Port Royal, 75005 Paris, France; 3Laboratory of Intestinal Immunity, INSERM UMR 1163, Institut Imagine, Paris, France and 4Université Paris Descartes-Sorbonne Paris Cité, 75006 Paris, France Correspondence: M. De Paepe ([email protected]) Received: 2 September 2019 Revised: 13 November 2019 Accepted: 5 December 2019 Published online: 6 January 2020

© The Author(s) 2020 New insights into intestinal phages R. Sausset et al. 206

Fig. 1 Phage life cycles. The production of new virions is realized either through lytic cycles for Caudovirales and Microviridae phages (left side of the figure, brown arrows) or through chronic infection in the case of filamentous phages, or Inoviridae (blue arrows). Both start with the recognition and infection of the targeted bacteria (1), followed by phage DNA replication and synthesis of new virions (2). In lytic cycles, new virions are released through bacterial lysis (3), while new virions of filamentous phages exit bacteria through a dedicated secretion apparatus, without bacterial lysis (4). Phages that reproduce only through lytic cycles are called virulent. By opposition, some phages, called temperate phages, in addition to performing either lytic or chronic cycles, are able to perform lysogenic cycles (pink arrows), whereby they enter a dormant state in the infected bacteria, the prophage state (5). The prophage, either integrated within the bacterial genome or in an episomal state, is replicated with the bacterial chromosome as long as bacteria divide (6). In some bacteria, generally when submitted to a stress, the prophage is induced and the phage resumes a lytic or a chronic cycle.

significant alignments to any known viral genome.13 Finally, This phenomenon, called pseudolysogeny, is poorly described, but intestinal phages are very challenging to cultivate, notably is suspected to exist in the intestinal environment (reviewed in because their bacterial hosts are mainly strict anaerobes that are ref. 21). Temperate phages, for their part, can use two very difficult to grow. However, starting from 0.2 or 0.45 µm filtered different lifestyles, the so-called “lysis–lysogeny choice”: infection fecal samples enriched in virions, shotgun deep sequencing has is either followed by a lytic cycle, as with virulent phages, or by permitted access to the human free-phage content (which will be lysogeny, whereby the phage enters a dormant state and is called designated below virome or phageome, since it comprises mainly a prophage. In this state, the expression of most phage genes is phages). Most phages appeared to be non-enveloped DNA repressed, preventing phage multiplication, but the phage viruses, either dsDNA Caudovirales or ssDNA Microviridae.In genome is replicated passively along with the bacterial genome. addition, a recent study indicates that ssDNA filamentous phages, The prophage can be either integrated in the bacterial chromo- or Inoviridae, that reproduce through chronic infection without some or extrachromosomal, like a plasmid. Following specific cues killing their host (Fig. 1) might also constitute a significant fraction described below, prophages can be activated, leading to phage of the human gut virome.18 In contrast, RNA phages were found to lytic cycle and death of the previously lysogenic bacteria. In be rare, if not completely absent, in the intestine.19,20 consequence, when considering intestinal phages, one should In the intestine as in other environments, phages can be take into account both free-phages and prophages. distinguished on the basis of their lifestyle, independently of Using recent assembly procedures, tens to thousands of DNA taxonomy (Fig. 1). Virulent phages essentially complete lytic phage contigs can be assembled per virome sample, depending cycles, whereby each infection is followed by virion production on sequencing depth. Yet, the functional roles of phages in the and host cell lysis. Yet, in some conditions such as nutritional gut ecosystem remain difficult to apprehend, notably because stress, virulent phage multiplication can be halted for a long time. they belong to entirely new and still uncharacterized genera and

Mucosal Immunology (2020) 13:205 – 215 New insights into intestinal phages R. Sausset et al. 207 ab

100 nm 1. CrAssphage 100 µm 2. Microviridae 3. Myoviridae 4. Siphoviridae

c 1.00 crAss-like Microviridae 0.75 Myoviridae Other Podoviridae Siphoviridae 0.50 Unassigned

Relative abundance 0.25

0.00 3 6 7 8 9 3 4 5 7 8 9 1 2 4 5 6 8 9 1 2 3 4 6 7 8 1 2 4 5 1 2 6 3 7 5 9 12 12 12 11 11 11 11 10 10 10 10 Fig. 2 Main intestinal phage types. a Epifluorescence microscopy image of a human fecal filtrate. Virus-like particles (VLPs) appear as bright dots following Sybr-gold staining of DNA (M. De Paepe, unpublished results). b Transmission electronic microscopy images of major intestinal phage types. Scale bars are 100 nm except for the Microviridae virion, for which it is 50 nm. 1: CrAssphage PhiCrAss001 (Podoviridae), reproduced from ref. 45 2: Microviridae virion isolated from surface water near a coastal aquaculture site, reproduced from ref. 126 3 and 4: Myoviridae and Siphoviridae virions respectively, extracted from human intestinal contents. Several Myoviridae virions are bound to a membrane vesicle. Reproduced from ref. 50 Copyright © 2014 Elsevier Masson SAS. c Inter-individual virome specificity and conservation in four human subjects over 12 months. Each phage cluster, which roughly corresponds to a phage genus, is outlined in black. Reproduced from ref. 29 even families, and their bacterial hosts are unknown. Several distinguish from virulent ones. For example, a study recent reviews have summarized the main characteristics of the focusing on the virome content of twins estimated the new findings in this emerging field.13,14,22 Their conclusions about proportion of temperate phages on the basis of the the phage community in the human gut are summarized below. proportion of phage contigs encoding an integrase gene, this gene being responsible for the integration of temperate (1) Most phages appear unique to each individual.23–25 The phage genome into the bacterial chromosome.23 However, existence of a “core phageome”, i.e., a small number of temperate phages do not necessarily encode an integrase phages shared between individuals, is under debate. A first gene, as many of them coexist with their host as study indicated that across 62 healthy individuals, 23 phage autonomously replicating episomes, and such method gives contigs (0.5% of all contigs) were shared by at least 50% of only a lower bound of the proportion of temperate phages. individuals and 132 contigs (3%) by at least 20% of The high abundance of temperate phages is compatible individuals.24 More recently, another study found that no with the observation that among isolated gut bacteria, most viral population (equivalent to viral species) was present in strains are lysogens and often polylysogens, notably within more than half of 132 samples from healthy individuals, and the dominant Bacteroidetes, Firmicutes, Actinobacteria, and that only 1% was shared by over 20% of individuals.24,25 Proteobacteria phyla.30–34 Several metagenomic studies These differences depend on the criteria used to identify a further suggest that the majority of intestinal bacteria carry given phage in a given virome sample and comparisons prophages.35,36 Yet, estimations of prophage prevalence suffer notably from the lack of consensus on phage based on genomic analyses cannot distinguish “active” taxonomy. Recent remarkable progresses in genome- prophages from “defective” ones, i.e., those that have lost based phage taxonomy26,27 may enable to better define the capacity to resume a lytic cycle due to mutations. the most common phage “types” in the human microbiota However, the presence of active prophages, which produce in the near future. virions, has been demonstrated experimentally in almost all (2) Healthy individuals tend to conserve the same phages over gut bacterial strain tested, suggesting that a significant time (tested over 1 year), especially the most abundant proportion of prophages detected by genomic analyses are ones23,28,29 (Fig. 2). Persistent phages, i.e., phages that reside active.33,34,37,38 High prevalence of temperate phages in at least one year in a given individual, seem also to be more human microbiota could participate in the stability of the commonly shared than others. Thus, among persistent phageome over time, as populations of lysogenic bacteria phages, 22 out of the 3639 viral clusters (0.6%; clusters constantly produce virions. might reflect genus level groupings) were shared by more (4) The identity of the hosts targeted by the phages is a crucial than half of the 10 individuals tested.29 but largely unanswered question. In 2016, Edwards et al.39 (3) In the human gut, temperate phages constitute at least 20% compared different methods for host prediction and those to 50 % of free phages, depending on samples but also on that gave the best results were simple nucleotide similarity the methodology, as temperate phages are difficult to searches between phage and bacterial genomes, either with

Mucosal Immunology (2020) 13:205 – 215 New insights into intestinal phages R. Sausset et al. 208 BLASTn or by searching the longest exact nucleotide match acids (Fig. 2a). However, if such nanoparticles are mainly virions, between a phage and a bacterial genome. With these some are membrane vesicles containing nucleic acid that are methods, the correct hosts were predicted for 37% and 40%, difficult to distinguish from virions by microscopy.47 In addition, respectively, of the 820 complete phages included in the immunologists often define VLPs as particles obtained by test. Several studies have used CRISPR spacers to predict spontaneous assembly of viral structural proteins, antigenically hosts,29,39–41 which led to very confident predictions but indistinguishable from infectious viruses, but that do not was restricted to hosts encoding CRISPR–Cas systems and in contain the viral genome.48 Despite these restrictions, we will which phage infection was relatively recent. Between 4% use indiscriminately virions and the microscopy-based definition and 13% of phages could be assigned to a host this way. of VLPs below. In most ecosystems, free phages outnumber Finally, the program WiSH bases its predictions on the bacteria by about tenfold.49 In contrast, the small number of similarity of the phage genome to that of its hosts. It uses a available studies enumerating virions in human stools indicates probabilistic approach that compares the composition in that in healthy subjects, bacteria outnumber virions. Two studies subsequences of nine nucleotides, or 9-mers, in phage and reported VLP abundances ranging from 1 × 108 to 2 × 109 VLPs/g bacterial genomes, and reaches good prediction even for in stools,50,51 whereas bacterial concentration is thought to be short 3 kb-long phage contigs, which is not the case for 1×1011 bacteria/g.52 Recently, the concentration of virions was other methods. Using the same large data set as Paez- also estimated by spiking viral metagenomic samples with a Espino et al.41, WiSH predicted a host at family level for 59% known concentration of a specific virus, indicating that total viral of the contigs.42 Using a mix of these approaches, recent loads may be between 2 × 108 and 8 × 1010 viral genome copies studies indicate that the spectrum of the bacterial hosts of per gram of stool.29 The ratio of free phages to bacteria that can the dominant phages reflects the microbiome composi- be extrapolated is thus comprised between 1 and 0.001, overall tion.29,43 Indeed, among 180 persistent phage clusters suggesting lower phage-induced mortality in the human gut identified, about one third could be linked to a bacterial microbiota than in other microbial ecosystems. An alternative genus, all of them belonging to abundant taxa, such as interpretation is that a large fraction of virions are “lost” in the Faecalibacterium and Bacteroides.29 GIT. Several factors may favor virion disappearance. First, virions (5) In individuals consuming a western diet, two particularly bind irreversibly to their bacterial receptors with a very fast rate prevalent and abundant phage taxa have been recently when receptor concentration is high.53 Such receptors can be described, both infecting Bacteroides species (Fig. 2): present on the surface of susceptible bacteria, but also on the surface of resistant bacteria or even on other structures such as Gokushovirinae are temperate phages belonging to the Micro- membrane vesicles or bacterial debris. In these last cases, viridae family, with a small circular ssDNA genome, and a small binding of virions on their receptor results in their inactivation. virion of 30 nm of diameter.31 Given the nature of their genome, AsthebacterialconcentrationintheGITismuchhigherthanin these virions can escape scrutiny or, on the contrary, be most other microbial ecosystems, a higher part of virions could overestimated depending on the kit and treatments used prior be lost by such a mechanism. For example, binding of virions on to sequencing (discussed in ref. 14). Therefore, their exact membrane vesicles is regularly reported in intestinal samples contribution to the phageome remains to be more accurately (Fig. 2b). Second, phages may be captured inside the mucus estimated. layer. Indeed, the Ig-like protein domain on the capsid of phage CrAss-like phages constitute a completely new clade of related T4 was shown to increase by fourfold virion binding to mucins phages present in at least 50% of individuals from western in vitro.54 As numerous intestinal phages possess similar motifs cohorts. They can represent up to 90% of the phageome of a on their capsid proteins,55 it may account for the large number single individual. CrAss-like phages have a dsDNA genome of of virions detected in the intestinal mucus.56 Whether these ~100 kb, a 90 nm diameter large head and a short tail (Fig. 2b). mechanisms operate in the human microbiota remains to be Their abundance has helped assembling many genomes of the investigated. Experiments in gnotoxenic mice (i.e. colonized clade, which presently span over four proposed subfamilies.44 One with a limited number of bacterial strains), in which the majority of them, CrAss001, infects Bacteroides intestinalis.45 Due to of the susceptible bacteria were killed by phage, suggest that difficulties in cultivating the CrAssphages, their lifestyle is not virion loss is highly dependent on individual phages considered. clearly established yet, but the presence of an integrase in some of In some reports, free phage to bacteria ratios over 100 were them suggests they may be temperate phages.29 observed in the feces of mice,37,57 whereas in others, ratios close to 1 were observed,58,59 suggesting either virion loss or very low viral amplification per infected bacteria, i.e., about one new In addition, a distinct prevalent phage taxon, called LAK phages, virion produced per lytic cycle. has recently been described in individuals from Tanzania and the region of Laksam, in Bangladesh. Their genomes were assembled directly from complete microbiota samples rather than using viral PHAGE–BACTERIUM INTERACTIONS IN THE GIT fractionation.46 Due to their remarkably large genome size (540 Phage–bacterium interactions in the gut microbiota are highly kb), they were coined “megaphages”, but their virions are not complex. First, phages often interact specifically with one single imaged yet. LAK phages have been shown to preferentially infect bacterial strain. Given bacterial diversity in the human gut, Prevotella, but their lifestyle is still unknown. hundreds to thousands of phage–bacteria pairs may potentially interact at any time. Second, as discussed below, the intestinal environment seems to protect a fraction of the genetically QUANTIFICATION OF PHAGES IN THE GIT susceptible bacterial populations from phage infection. Third, Defining the number of phages present in the GIT should help to both antagonistic and mutualist interactions are possible. For predict their impact on bacteria. As already stressed, both free example, bacteria can be killed by phages, either by infection phages and prophages should be taken into account. Prophages followed by lytic cycle (predation) or upon prophage induction, are present in the majority of bacterial genomes and are thus but on the contrary, as we will see below, prophages and approximately as numerous as bacteria in the GIT. Free phages, filamentous phages can provide benefits to their host. Finally, due produced during lytic cycles, are generally enumerated as virus- to the high evolutionary capacities of phages and bacteria, like particles (VLPs). VLPs are nanoparticles that can be observed phage–bacteria interactions may change very rapidly over time. by epifluorescence microscopy following staining of nucleic Below we discuss the phage–bacterium interactions that have

Mucosal Immunology (2020) 13:205 – 215 New insights into intestinal phages R. Sausset et al. 209

abT4 phage - E. coli B40-8 phage - B. fragilis 9 11

2. Bacteria in spatial refuges 7 9 • biofilms [bacteria or phage]/g

10 • crypts • mucus layer Log 5 7 0 10 20 30 40 0 10 20 30 40 1. Phenotypic F1 phage - C. sporogenes VD13 phage - E. faecalis variability in receptor 9 expression and/or 10 accessibility

8 7 [bacteria or phage]/g 10 6 4. Poor phage diffusion

Log 3. Free-phage loss by due to fibers from food 4 5 adsorption (membrane or mucus 0 10 20 30 40 0 10 20 30 40 vesicles, debris, non- Time (d) Time (d) growing bacteria …) Fig. 3 Phage-bacterium interactions in the GIT. a Illustration of the variability in the populations of phages and targeted bacteria in feces of gnotobiotic mice, reproduced from ref. 57 Germ-free mice were colonized with ten bacterial strains representing the major phyla of the human gut microbiota (Clostridium sporogenes, E. faecalis, Bacteroides fragilis, B. ovatus, B. vulgatus, Parabacteroides distasonis, Klebsiella oxytoca, Proteus mirabilis, E. coli Nissle 1917, and Akkermansia muciniphila). Dotted lines indicate the time of addition of 2 × 106 PFU of a virulent phage targeting one bacterial strain exclusively. Bacteria and phages were quantified by quantitative PCR (blue and red lines, respectively). In the case of E. coli, after a rapid drop in bacterial population after T4 phage administration, phage and bacteria coexist without selection of phage-resistant mutants. In the cases of C. sporogenes and B. fragilis, bacterial populations are transiently affected by phage administration, and phages reach very high concentrations. Finally, in the case of E. faecalis, the bacterial population drops and does not recover its initial level over time, due to a defect of colonization by phage-resistant mutants. b Hypotheses for the co-existence of phage and susceptible bacteria in the GIT. (1) Important bacterial phenotypic variability, e.g., in phage receptor expression or growth rate, would render some bacteria phage-susceptible, whereas other genetically identical bacteria would be resistant. (2) Spatial heterogeneity of the environment (such as intestinal crypts, biofilm- like structures on food debris, or the inner part of the mucus layer) would create a refuge for bacteria and prevent access to phages. (3) Important loss of virions due to specific and nonspecific adsorption would sufficiently lower free-phage concentration to protect bacteria. (4) The presence of numerous long chain carbohydrates fibers from food and mucus would hinder phage diffusion, preventing them to adsorb on bacteria. None of these mechanisms has been formally demonstrated in the GIT. been demonstrated in the GIT and their impact on the gut bacterial concentration.67,68 Altogether, these results suggest that microbiota composition. in the GIT, prophage induction might constitute a significant burden for their bacterial host, and alter microbiota composition. Prophage induction Indeed, it has been shown in mouse models that a high prophage Most prophages are highly stable, but environmental stressors or induction rate modifies the equilibrium between bacterial strains stochastic fluctuations in phage repressor concentrations can by disfavoring lysogens.38,58 trigger their induction, i.e., resumption of the lytic cycle and subsequent lysis of the host bacteria. In general, cellular signals Lysogenic conversion triggering prophage induction are DNA damages, via the Some prophages compensate their detrimental effects on their destabilization of the repressor or master regulator of lysogeny. host by providing them beneficial traits that augment their fitness Quinolone antibiotics, which cause DNA double-strand breaks, and can confer completely novel phenotypes, a phenomenon are the most described prophage inducers in the GIT.60 In known as “lysogenic conversion.” Examples of lysogenic conver- consequence, as E. coli stx prophages encode the shiga toxin, sion are numerous and recently reviewed by Taylor et al.69 and treatment of human shigatoxigenic E. coli infection with Wahl et al.70 in Salmonella. These new phenotypes comprise quinolones have significant adverse clinical consequences.60 immunity to phage super-infection, resistance to other phages, Spontaneous induction rates, estimated in vitro to be between tolerance to various stresses, pathogenicity, and, very rarely, 10−7 and 10−4, were generally considered to be too low to antibiotic resistance. With respect to antibiotic resistance genes negatively impact the lysogen’s fitness (reviewed in ref. 61). Yet, (ARGs), their presence within phage genomes is debated. Most several pieces of evidence suggest that induction rates are reports are flawed by either excessive bacterial DNA contamina- globally higher in the murine GIT than in classical in vitro growth tion of the virome samples71–74 or inappropriate thresholds used cultures, due to more frequent activation of the DNA damage for the similarity search against ARG databases.75 When both response (SOS response).58,62,63 In the case of Lactobacillus reuteri, pitfalls were eliminated, ARG were rare in phage genomes (around SOS activation was proposed to result from the activation of or <1 gene among 105 genes analyzed, on phage genomes and specific bacterial metabolic pathways in the GIT.62 Conflicting phage metagenomic contigs respectively),76 confirming an results were reported concerning the possible increase or, on the ancient observation.77 Further illustrating the potential of contrary, decrease of E. coli stx prophage induction rates by gut prophages to confer fitness advantages to their bacterial hosts, metabolites such as nitric oxide or bile salts (reviewed in ref. 64). a very recent study analyzing sponge-associated virome sug- Bile salts were also shown to induce some Salmonella pro- gested that some prophages encoding ankyrin-repeat proteins phages.65. Besides, intestinal inflammation increases the induction may provide anti-inflammatory properties to their bacterial host of a Salmonella prophage in mice.66 Finally, one has to mention and foster host–microbe symbiosis.78 the recent discovery of quorum-sensing pathways in phages of Independently of lysogenic conversion, a few prophages can Enterococcus faecalis and Vibrio cholerae, suggesting that some function as genetic switches, a phenomenon known as “active prophages could regulate their induction rate in response to lysogeny” (reviewed in ref. 79). In active lysogeny, prophage

Mucosal Immunology (2020) 13:205 – 215 New insights into intestinal phages R. Sausset et al. 210 excision does not result in lytic replication, but restores the conferring phage resistance to V. cholerae occurred and were integrity of the bacterial gene or operon in which the prophage selected in two patients during cholera infection. In addition, was integrated. For the intracellular pathogen Listeria monocyto- metagenomic studies conducted on conventional animal’s feces genes, active lysogeny was shown to facilitate bacterial escape showed a correlation between increases in specific phages and from the phagosomes.80 In a third situation, the production of reductions in particular bacterial taxa, but without definitive virions itself benefits the lysogen population (reviewed in ref. 61). proofs of direct interactions.35,95 For example, production of filamentous phage particles by Lastly, an interesting but poorly investigated phage–bacteria Neisseria meningitidis promotes bacterial aggregation in vitro via antagonistic interaction in the GIT is the potential ability of some the formation of bundles of phage filaments, supposedly phages to infect simultaneously several bacterial strains belong- increasing the bacterial colonization of the nasopharynx.81 ing, or not, to the same species. Although a large amount of work Due to the beneficial traits conferred to bacteria by prophages, has indicated that most phages infect only a few strains of the temperate phages are often considered as mutualists rather than same bacterial species, several clues suggest that intestinal parasites of their bacterial host, and the prevalence of lysogeny is phages might be more promiscuous than usually assumed. First, often interpreted as an evidence that prophages increase the a phage infecting Faecalibacterium prausnitzii was shown to also selective fitness of their hosts (reviewed in ref. 82). However, if infect Blautia hansenii, a species only distantly related to benefits conferred by prophages to pathogenic bacteria are well F. prausnitzii.33 Second, metagenomic analyses of CRISPR spacers demonstrated, to the best of our knowledge, there is no direct have suggested that some phages may have a broad spectrum of demonstration that lysogenic conversion increases the coloniza- hosts in the human gut.29 In addition, phage evolution resulting in tion ability of a gut commensal in vivo, except very transiently by the ability to infect a new host (host jump) was reported in a phage killing of a closely related strain.58,83 Alternatively, the mouse model of coevolution in the GIT.89 prevalence of lysogeny may reflect regular lysogenization of In conclusion, phage–bacteria interactions in the GIT appear to bacteria by temperate phages, as evidenced in the GIT of be highly diverse, involving partial bacterial resistance, fast animals,58,66,84,85 and even in the human gut microbiota.86 evolution of phages and bacteria, lysogenization and probably indirect cascading effects on several bacterial species and their Predation phages. A large body of information on phage predation in the GIT comes from animals with a modified microbiota, either gnotobiotic or treated by antibiotics, and colonized with specific bacterial strains IMPACT OF PHAGES ON HOST THROUGH GUT MICROBIOTA and their associated phages. In all cases, phages multiplied MODULATION successfully in the GIT. Yet, the decrease in the targeted bacteria is Phage therapy of intestinal disease infections very variable (Fig. 3a).57,59,87,88 In some experiments, phage- The complex relationship between phages and bacteria probably resistant bacterial mutants were observed. However, and in explains the relative unpredictability of phage therapy trials contrast with results obtained in vitro, these mutants never against intestinal pathogens. Phage therapy was nearly aban- entirely replaced the susceptible bacteria.37,57,89 In other cases, doned in most western countries 60 years ago given its phage-mediated bacterial mortality was relatively limited and unpredictable outcomes and the efficacy of newly discovered neither depletion of the bacterial population, nor selection of antibiotics. The compassionate use of phage therapy has been phage-resistant mutants were observed.57,59,87,88 Overall, these extensively practiced, though, in several countries, notably in data suggest that the intestinal environment provides bacteria Poland. The Eliava Institute, in Georgia, has been routinely treating with spatial refuges or alternative resistance mechanisms, such as various infections, and notably gut infections, with phage cocktails phenotypic resistance (Fig. 3b). For example, bile salts were shown for more than a century, but no evaluation is available. An to inhibit the infection of E. coli by several phages in vitro, possibly evaluation of Polish treatments reported symptomatic ameliora- via the repression of a phase-variable cell-surface protein.90 tion and clinical recovery in 40 and 18% of patients, respectively, Variability in phage resistance can also result from spatial which is quite encouraging for patients who previously failed to heterogeneity in the bacterial population driven by gradients of respond to antibiotic treatments.96 The mounting incidence of abiotic factors, such as pH and oxygen, but also of molecules such severe infections by multi-drug resistant bacteria, along with as mucins, bile acids, and short-chain fatty acids. These gradients recent progresses in phage biology has prompted re-evaluation of can modify the physiology of bacteria and, consequently, their phage therapy. A randomized controlled clinical trial was recently susceptibility to phages, as shown ex vivo in Maura et al.91 Fluxes conducted against E. coli diarrhea in Bangladesh. It was may also prevent phages to access some compartments, such as unfortunately unsuccessful. Yet, it was subsequently shown that intestinal crypts.15,91 Finally, the crowded environment of the GIT E. coli was probably not the cause of the diarrhea.97,98 This large may hamper phage diffusion (reviewed in refs. 92 and 93). Overall, scale study conducted up to stage III nevertheless permitted to most current reports point to a transient effect of phages on the establish the innocuous character of phage cocktails for humans.98 population sizes of targeted bacteria in the GIT microbiota of Besides classical phage therapy with defined phage mixtures, animal models. Nevertheless, successful phage therapy trials transfer of fecal filtrates containing phages but no bacteria, was realized before the 70s against intestinal bacteria indicate that proposed as an alternative to fecal microbiota transplantation phages may efficiently reduce bacterial loads in the human GIT.5–7 (FMT) for Clostridium difficile infection.99 The transfer of fecal Another related question is the importance of phage-mediated filtrates to five patients with chronic C. difficile infection was mortality in a “natural” microbiota, i.e., without the artificial successful in all patients, suggesting that phages present in the ingestion of high phage doses. At the timescale of evolution, fecal filtrates mediate many of the effects of FMT, but the phage killing left traces in the genomes of intestinal bacteria in mechanisms remain unknown.99 the form of a large repertoire of anti-phage systems and highly diverse CRISPR spacer sequences.40 However, such traces of Impact of phages on individuals with a healthy microbiota phage-mediated selection could result from a low level of phage- Beyond phage therapy, given the numerous interactions between induced mortality, as even a small selective advantage can lead to intestinal bacteria and the host, it is important to evaluate the allele fixation if sufficient time is given. At a shorter ecological extent of phage-induced modifications of microbiota functionality. timescale, there are also several evidences of significant natural First, as discussed previously, prophages modify numerous phage-mediated bacterial mortality in humans and animals. First, bacterial phenotypes, and notably virulence, thereby impacting a study by Seed et al.94 strongly suggests that mutations host–bacteria interactions69,70,78. In particular, the ankyphages

Mucosal Immunology (2020) 13:205 – 215 New insights into intestinal phages R. Sausset et al. 211 recently discovered in sponges might contribute to promote One study notably reported a global shift in virome composi- bacterial protection from the eukaryotic immune system.100 tion in patients with inflammatory bowel disease (IBD), either Indeed, transformed E. coli bacteria expressing a phage ankyrin Crohn’s disease or ulcerative colitis, compared with household protein better resisted murine bone-marrow-derived macrophage controls.115 Whether such shift implies that phages are actively phagocytosis and induced a diminished inflammatory cytokine involved in disease, or simply reflects the shift in microbiota response in these macrophages. Interestingly, phage contigs composition remains to be shown. Nonetheless, re-analysis of encoding similar ankyrin proteins were found in various host- thedatafocusingonthetemperatephagesofF. prausnitzii,a associated environments, including human oral and gut viromes, species generally depleted in IBD patients, showed that two F. suggesting some wide-range phenomenon.100 prausnitzii phages were significantly more prevalent in IBD Second, beyond lysogeny, phages can impact the mammalian samples and two others were significantly more abundant,33 host physiology through microbiota modulation, by depleting suggesting enhanced temperate phage-mediated mortality bacterial species important for homeostasis. However, as of F. prausnitzii in IBD. One may speculate that the inflamma- discussed previously, in animal models, phages never fully tory environment of the gut might increase prophage eradicate bacterial species in the gut, due to partial protection induction, as shown in Salmonella,66 thereby promoting of bacteria and fast evolution of resistant bacterial mutants. bacterial lysis and aggravating dysbiosis, and reinforcing the Effects of phages on microbiota composition at the species level inflammatory loop. are therefore expected to be transient, unless they act in conjunction with other factors, such as other members of the microbiota. Studies aiming at detecting cascading effects on bacteria not targeted by the phage introduced have brought Lumen contrasting results. Studies in gnotobiotic rats and mice showed that beyond phage-mediated targeted knockdown of suscep- Mucus tible species in the gut, the concentration of non-targeted bacteria could be impacted, notably through modification of Phage bacterial interactions.57,101 However, other studies in conven- antigens tional animals showed no modification of the non-targeted microbiota.102,103 Similarly, a modification of the human micro- 1 biota composition has been observed upon oral administration 104 of phages in one study, but not in another involving M cell children.105 Such differences could be related to the difficulty of attributing shifts in microbiota composition to phages or to its natural temporal variability. Phages could also impact indirectly the host if phage resistance affects the ability of bacteria to colonize the gut and to interact with their host. Indeed, many bacterial structures used as phage Subepithelial receptors, such as lipopolysaccharides, porins, flagella, fimbriae 2 and pili, play a key role in bacterial colonization, growth, compartment persistence, virulence and recognition by the host's immune system in the GIT (reviewed in refs. 2,16). For example, mutations in porins such as TonB or LamB, conferring phage resistance in E. coli, can diminish the nutritional competence of the bacteria, possibly affecting its gut colonizing ability. In particular, phage resistance B cell 3 was shown to negatively impact the ability of bacteria to develop IgA antibiotic resistance, as bacterial mutants resistant to phage were (gut) Dendritic more sensitive to antibiotics. For example, mutations in Pseudo- IgG cell monas conferring in vitro phage resistance affected a porin (serum) fl 106 involved in multi-drug ef ux, whereas in the case of E. faecalis, 4 epa mutations changing its exopolysaccharide composition rendered the strain more sensitive to cell-wall-targeting antibiotics + in mice.107 Moreover, several E. faecalis epa mutations could CD4 107,108 T cell simultaneously confer phage resistance and defective γ intestinal colonization in mice.109 Indeed, a promising type of TLR-9 IFN- phage therapy consists in selecting phages on their faculty to favor the emergence of phage-resistant bacterial mutants that 110 fi Fig. 4 Potential interactions between phages, epithelial cells, and become antibiotic sensitive. Phages may also indirectly bene t host immune cells in the intestine. Phage tropism for the mucus their mammalian host by increasing the diversity of microbial 16,111,112 may promote penetration of phages within the body through communities, a known factor of stability of the gut endocytosis and transcytosis in intestinal epithelial cells (1), or microbiota.113 Finally, some phages can perform generalized through sampling by dendritic cells (2). Dendritic cells endocytose transduction, permitting the transfer of bacterial DNA from an phages, may be captured in the intestinal lumen via extended infected cell to another. Some data indicate that a substantial dendrites, or exocytosed in the subepithelial compartment. Once fraction of both temperate and virulent phages can perform such endocytosed, phage nucleic acids can trigger TLR pathways, notably transduction (reviewed in refs. 114), so that this phenomenon may TLR9-dependent pathways (3), and stimulate adaptive immune take place regularly in the GIT microbiota, participating in bacterial responses (4). Although mechanisms of B and T-cell activation by phages are not fully elucidated, recent studies showed that evolution. activation of B cells leads to the secretion of phage-specific antibodies, both in the intestine and in the systemic compartment. A role of phages on dysbioses? In addition, activation of T cells in the Peyer’s patches and Several studies have investigated correlations between virome mesenteric lymph nodes results in production of cytokines, such composition and the physio-pathological state of individuals. as IFN-γ.

Mucosal Immunology (2020) 13:205 – 215 New insights into intestinal phages R. Sausset et al. 212 INTERACTIONS BETWEEN INTESTINAL PHAGES AND THE respond than in those who responded to this treatment.12 A MAMMALIAN IMMUNE SYSTEM second recent study provided evidence of TLR3-triggering by a In addition to their indirect impact on hosts immune responses filamentous phage present in the P. aeruginosa strains infecting through changes in microbiota, phages may also interact directly chronic human wounds.124 Activation of TLR3 required endocy- with the host immune system and trigger immune responses tosis of phages by immune cells and, much surprisingly, neo- (Fig. 4). Whether intestinal phages disseminate outside the gut synthesis of phage RNA into the immune cell, which resulted in lumen and interact with immune cells in the intestinal mucosa or the production of type I IFN. This cytokine, in turn, inhibited the at distant sites is a first matter of debate. Some studies in rodents production of TNF by macrophages, thereby impairing phagocy- and humans stated that no or few phages were detected in the tosis and bacterial clearance, and delaying wound healing.124 bloodstream after oral administration. In contrast, others demon- Interestingly, this study suggests that phage uptake by mamma- strated effective dissemination, largely correlating with the dose lian immune cells, rather than simple cell-surface interactions, may of ingested phages116 (and reviewed in refs. 11,12). In addition, be necessary to trigger host immune responses. phage-neutralizing antibodies against naturally occurring intest- Overall, a limited set of data suggests that phages may exert a inal phages have been detected in the sera of different direct effect on immune cells. More work is however needed to mammalian species, indicating phage contact with host immune define whether the effects observed in vitro or in vivo upon oral cells (reviewed in ref. 17). A recent study further showed that gavage with large amounts of purified virions can be recapitulated repeated oral exposure to high doses of the E. coli phage T4 in more physiological conditions in the gut and how these effects induced simultaneously a specific protective secretory IgA may interfere or synergize with the changes induced by phages in response in the gut and a serum IgG response.116 Overall, these the composition of the microbiota to impact mammalian health data suggest that phages can translocate through the mouse gut and disease, both in and out of the intestine. epithelium and activate the host immune system both locally and systemically. Translocation may occur across epithelial cells. Accordingly, recent in vitro data showed oriented phage CONCLUSIONS transcytosis across several epithelial cell layers.117 Binding of During the last decade, viral metagenomics has shed light on the phages to mucin glycoproteins54 may further promote contact taxonomic composition and dynamics of the viral component of with epithelial cells and transcytosis. Alternatively, phage uptake the gut microbiota. Deep sequencing and novel assembly may involve dendritic cells. Indeed, dendritic cells exposed in vitro methods have allowed the description of completely new phages. to phage particles were shown to efficiently phagocytose these These approaches have notably revealed that the virome particles118–120 and dendritic cells are thought to extend dendrites composition is highly variable, with only a small fraction of through the intestinal epithelium and thereby sample luminal phages shared among individuals. Not surprisingly, the most bacteria and particles.121 However, the incidence of this phenom- abundant viruses were found to infect Bacteroides and Clostridiales enon seems rare in the steady state, suggesting that luminal species that are dominant in the microbiota. In parallel, phage uptake by dendritic cells may be a relatively rare event. experiments in animals with a simplified microbiota allowed to Although more work is required to establish how phages can explore phage–bacteria antagonistic interactions in the gut and get access to host immune cells in vivo, recent studies have have uncovered variable outcomes. For some phage–bacteria analyzed their interactions with host intracellular immune path- pairs, the intestinal environment somehow protects bacteria and ways and their capacity to trigger immune responses in vitro and only a small fraction of genetically susceptible bacteria are killed in vivo. Following the unexpected observation that the treatment by their specific phages. In other cases, most of the phage- of mice with a cocktail of antiviral drugs aggravated dextran susceptible bacterial population is replaced by resistant mutants sulfate sodium (DSS)-induced colitis, Yang et al.122 suggested that within a few days, indicating very efficient phage infection. gut resident viruses, a large fraction of which being phages, Therefore, which outcome will prevail in complex natural protect against intestinal inflammation via a mechanism involving microbiota remains difficult to predict. Traces of phage predation TLR3 and TLR7-dependent production of interferon (IFN)-β. In this can be found in gut bacterial genomes, but these traces could study, it is however difficult to assign a role to phages in TLR result from a low level of phage-induced mortality. A low level of triggering as the antiviral drugs used target only eukaryotic phage predation could nevertheless be crucial to shape micro- viruses. Moreover, antiviral treatment was associated with changes biota composition and functionality, by affecting bacterial in the microbiota, which may have also contributed to stimulate evolution through horizontal gene transfer, but also by promoting protection. Three other studies suggest that phages also directly bacterial diversity. At shorter ecological timescales, the impact of interact with innate immune cells. Van Belleghem et al.123 phages is less well established. In particular, the role of phages in observed that in vitro incubation of peripheral blood monocytes the dysbioses that accompany various pathological conditions with purified Staphylococcus aureus or Pseudomonas aeruginosa remains poorly defined. Thanks to recent progresses in the phages induced a transcriptional response in monocytes, and determination of phage–bacteria pairs, longitudinal studies can notably enhanced the transcription of interleukin (IL)-1, IL-6, and now be undertaken to identify possible relationships between tumor necrosis factor (TNF). Strengthening this in vitro study, temporal shifts in bacteria and their associated phages and to Gogokhia et al.12 observed the expansion of IFN-γ-producing delineate whether phages may contribute to dysbiosis and CD4+ T cells and of CD8+ T cells in the Peyer’s patches of germ- disease or, on the contrary, help to maintain microbiota stability free mice orally treated with a lipopolysaccharide-free fraction of by preserving bacterial diversity. purified E. coli phages. They further showed that dendritic cells Further studies are also needed to substantiate possible direct incubated with phages or phage-derived DNA could produce interactions of phages with immune cells and to define whether several cytokines including IL-12, IL-6, and IL-10, and stimulate and how such direct effects may modify the composition of the CD4+ T-cell production of IFN-γ through a TLR9-dependent, but bacterial microbiota and influence host health or disease. Defining TLR3-independent, signal. In contrast with the effect of eukaryotic the mechanisms that determine the outcome of phage–bacteria viruses reported by Yang et al.122 oral administration of the phage interactions in the gut is particularly instrumental in the cocktail aggravated DSS colitis in a TLR9- and IFN-γ-dependent perspective of phage therapy, which, to cite Brüssow,125 “is manner.12 Noticeably, the authors made a parallel with their without doubt an interesting approach to the antibiotic resistance observations in patients with ulcerative colitis treated by fecal problem and merits intensified research to get out of the fruitless transplantation, showing that the relative intestinal abundance of confrontation between enthusiasm from the East and lingering Caudovirales bacteriophages was higher in patients who failed to Western skepticism”.

Mucosal Immunology (2020) 13:205 – 215 New insights into intestinal phages R. Sausset et al. 213 AUTHOR CONTRIBUTIONS 30. Touchon, M., Bernheim, A. & Rocha, E. P. Genetic and life-history traits associated All authors participated to the writing of the review. with the distribution of prophages in bacteria. ISME J. 10, 2744–2754 (2016). 31. Krupovic, M. & Forterre, P. Microviridae goes temperate: microvirus-related proviruses reside in the genomes of Bacteroidetes. PLoS ONE 6, e19893 (2011). ADDITIONAL INFORMATION 32. Bobay, L. M., Rocha, E. P. & Touchon, M. The adaptation of temperate bacter- – Competing interests: The authors declare no competing interests. iophages to their host genomes. Mol. Biol. Evol. 30, 737 751 (2013). 33. Cornuault, J. K. et al. Phages infecting Faecalibacterium prausnitzii belong to Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims novel viral genera that help to decipher intestinal viromes. Microbiome 6,65 in published maps and institutional affiliations. (2018). 34. Lugli, G. A. et al. Prophages of the genus Bifidobacterium as modulating agents of the infant gut microbiota. Environ. Microbiol. 18, 2196–2213 (2016). 35. Waller, A. S. et al. Classification and quantification of bacteriophage taxa in REFERENCES human gut metagenomes. ISME J. 8, 1391–1402 (2014). 1. Nobrega, F. L. et al. Targeting mechanisms of tailed bacteriophages. Nat. Rev. 36. Kim MSB, J. W. Lysogeny is prevalent and widely distributed in the murine gut Microbiol 16, 760–773 (2018). microbiota. ISME J. 12, 1127–1141 (2018). 2. Dy, R. L., Richter, C., Salmond, G. P. & Fineran, P. C. Remarkable mechanisms in 37. Cornuault J. K., et al. The enemy from within: a prophage of Roseburia intestinalis microbes to resist phage infections. Annu. Rev. Virol. 1, 307–331 (2014). systematically turns lytic in the mouse gut, driving bacterial adaptation by 3. Ofir GS, R. Contemporary phage biology: from classic models to new insights. CRISPR spacer acquisition. ISME J. https://doi.org/10.1038/s41396-019-0566-x Cell 172, 1260–1270 (2018). (2019). 4. Twort, F. W. An investigation on the nature of ultra-microscopic viruses. Lancet 38. Oh, J. H. et al. Prophages in Lactobacillus reuteri are associated with fitness trade- 2, 1241–1243 (1915). offs but can increase competitiveness in the gut ecosystem. Appl. Environ. 5. d'Hérelle, F. On an invisible microbe antagonistic toward dysenteric bacilli. Microbiol. 86, e01922–19 (2019). pii. Compte Rendu de. l'Acad.émie des. Sci. 165, 373–375 (1917). 39. Edwards, R. A., McNair, K., Faust, K., Raes, J. & Dutilh, B. E. Computational 6. d'Hérelle, F. Studies upon Asiatic cholera. Yale J. Biol. Med 1, 195–219 (1929). approaches to predict bacteriophage-host relationships. FEMS Microbiol. Rev. 40, 7. Babalova, E. G. et al. [Preventive value of dried dysentery bacteriophage]. Zh. 258–272 (2016). Mikrobiol. Epidemiol. Immunobiol. 45, 143–145 (1968). 40. Stern, A., Mick, E., Tirosh, I., Sagy, O. & Sorek, R. CRISPR targeting reveals a 8. Kutter, E. et al. Phage therapy in clinical practice: treatment of human infections. reservoir of common phages associated with the human gut microbiome. Curr. Pharm. Biotechnol. 11,69–86 (2010). Genome Res 22, 1985–1994 (2012). 9. Weinbauer, M. G. Ecology of prokaryotic viruses. FEMS Microbiol Rev. 28, 41. Paez-Espino, D. et al. Uncovering Earth’s virome. Nature 536, 425–430 (2016). 127–181 (2004). 42. Galiez, C., Siebert, M., Enault, F., Vincent, J. & Soding, J. WIsH: who is the host? 10. Chatterjee, A. & Duerkop, B. A. Beyond bacteria: bacteriophage-eukaryotic host Predicting prokaryotic hosts from metagenomic phage contigs. Bioinformatics interactions reveal emerging paradigms of health and disease. Front Microbiol 9, 33, 3113–3114 (2017). 1394 (2018). 43. Moreno-Gallego, J. L. et al. Virome diversity correlates with intestinal micro- 11. Keen, E. C. & Dantas, G. Close encounters of three kinds: bacteriophages, biome diversity in adult monozygotic twins. Cell Host Microbe 25, 261–272 e265 commensal bacteria, and host immunity. Trends Microbiol. 26, 943–954 (2018). (2019). 12. Gogokhia, L. et al. Expansion of bacteriophages is linked to aggravated intestinal 44. Guerin, E. et al. Biology and taxonomy of crAss-like bacteriophages, the inflammation and colitis. Cell Host Microbe 25, 285–299 e288 (2019). most abundant virus in the human gut. Cell Host Microbe 24, 653–664 e656 13. Aggarwala, V., Liang, G. & Bushman, F. D. Viral communities of the human gut: (2018). metagenomic analysis of composition and dynamics. Mob. DNA 8, 12 (2017). 45. Shkoporov, A. N. et al. PhiCrAss001 represents the most abundant bacterioph- 14. Shkoporov, A. N. & Hill, C. Bacteriophages of the human gut: the “known age family in the human gut and infects Bacteroides intestinalis. Nat. Commun. 9, unknown” of the microbiome. Cell Host Microbe 25, 195–209 (2019). 4781 (2018). 15. De Sordi, L., Lourenco, M. & Debarbieux, L. The battle within: interactions of 46. Devoto, A. E. et al. Megaphages infect Prevotella and variants are widespread in bacteriophages and bacteria in the gastrointestinal tract. Cell Host Microbe 25, gut microbiomes. Nat. Microbiol. 4, 693–700 (2019). 210–218 (2019). 47. Forterre, P., Soler, N., Krupovic, M., Marguet, E. & Ackermann, H. W. Fake virus 16. Scanlan, P. D. Bacteria-bacteriophage coevolution in the human hut: implica- particles generated by fluorescence microscopy. Trends Microbiol. 21,1–5 tions for microbial diversity and functionality. Trends Microbiol. 25, 614–623 (2013). (2017). 48. Chackerian, B. Virus-like particles: flexible platforms for vaccine development. 17. Van Belleghem J. D., Dabrowska K., Vaneechoutte M., Barr J. J., Bollyky P. L. Expert Rev. Vaccines 6, 381–390 (2007). Interactions between bacteriophage, bacteria, and the mammalian immune 49. Cobian Guemes, A. G. et al. Viruses as winners in the game of life. Annu. Rev. system. Viruses 11, pii: E10 (2018). Virol. 3, 197–214 (2016). 18. Roux, S. et al. Cryptic inoviruses revealed as pervasive in bacteria and archaea 50. Hoyles, L. et al. Characterization of virus-like particles associated with the human across Earth’s biomes. Nat. Microbiol. 4(11), 1895–1906 (2019). faecal and caecal microbiota. Res. Microbiol. 165, 803–812 (2014). 19. Zhang, T. et al. RNA viral community in human feces: prevalence of plant 51. Kim, M. S., Park, E. J., Roh, S. W. & Bae, J. W. Diversity and abundance of single- pathogenic viruses. PLoS Biol. 4, e3 (2006). stranded DNA viruses in human feces. Appl. Environ. Microbiol. 77, 8062–8070 20. Lim, E. S. et al. Early life dynamics of the human gut virome and bacterial (2011). microbiome in infants. Nat. Med. 21, 1228–1234 (2015). 52. Sender, R., Fuchs, S. & Milo, R. Revised estimates for the number of human and 21. Los, M. W. & Wegrzyn, G. Pseudolysogeny. Adv. Virus Res. 82, 339–348 (2012). bacteria cells in the body. PLoS Biol. 14, e1002533 (2016). 22. Manrique, P., Dills, M. & Young, M. J. The human gut phage community and its 53. Berg, H. C. & Purcell, E. M. Physics of chemoreception. Biophys. J. 20, 193–219 implications for health and disease. Viruses 9, E141 (2017). (1977). 23. Reyes, A. et al. Viruses in the faecal microbiota of monozygotic twins and their 54. Barr, J. J. et al. Bacteriophage adhering to mucus provide a non-host-derived mothers. Nature 466, 334–338 (2010). immunity. Proc. Natl Acad. Sci. USA 110, 10771–10776 (2013). 24. Manrique, P. et al. Healthy human gut phageome. Proc. Natl Acad. Sci. USA 113, 55. Fraser, J. S., Yu, Z., Maxwell, K. L. & Davidson, A. R. Ig-like domains on bacter- 10400–10405 (2016). iophages: a tale of promiscuity and deceit. J. Mol. Biol. 359, 496–507 (2006). 25. Gregory A. C., Zablocki O., Howell A., Bolduc B., Sullivan M. B. The human gut 56. Lepage, P. et al. Dysbiosis in inflammatory bowel disease: a role for bacter- virome database. Cold Spring Harb. Lab. https://doi.org/10.1101/655910 (2019). iophages? Gut 57, 424–425 (2008). 26. Low, S. J., Dzunkova, M., Chaumeil, P. A., Parks, D. H. & Hugenholtz, P. Evaluation 57. Hsu, B. B. et al. Dynamic modulation of the gut microbiota and metabolome by of a concatenated protein phylogeny for classification of tailed double-stranded bacteriophages in a mouse model. Cell Host Microbe 25, 803–814 e805 (2019). DNA viruses belonging to the order Caudovirales. Nat. Microbiol. 4, 1306–1315 58. De Paepe, M. et al. Carriage of lambda latent virus is costly for its bacterial host (2019). due to frequent reactivation in monoxenic mouse intestine. PLoS Genet. 12, 27. Barylski, J. et al. Analysis of spounaviruses as a case study for the overdue e1005861 (2016). reclassification of tailed phages. Syst. Biol. 69, 110–123 (2019). 59. Weiss, M. et al. In vivo replication of T4 and T7 bacteriophages in germ-free mice 28. Minot, S. et al. The human gut virome: inter-individual variation and dynamic colonized with Escherichia coli. Virology 393,16–23 (2009). response to diet. Genome Res. 21, 1616–1625 (2011). 60. Zhang, X. et al. Quinolone antibiotics induce Shiga toxin-encoding bacter- 29. Shkoporov, A. N. et al. The human gut virome is highly diverse, stable, and iophages, toxin production, and death in mice. J. Infect. Dis. 181, 664–670 individual specific. Cell Host Microbe 26, 527–541 e525 (2019). (2000).

Mucosal Immunology (2020) 13:205 – 215 New insights into intestinal phages R. Sausset et al. 214 61. Nanda, A. M., Thormann, K. & Frunzke, J. Impact of spontaneous prophage 91. Maura, D., Galtier, M., Le Bouguenec, C. & Debarbieux, L. Virulent bacteriophages induction on the fitness of bacterial populations and host-microbe interactions. can target O104:H4 enteroaggregative Escherichia coli in the mouse intestine. J. Bacteriol. 197, 410–419 (2015). Antimicrob. Agents Chemother. 56, 6235–6242 (2012). 62. Oh, J. H. et al. Dietary fructose and microbiota-derived short-chain fatty acids 92. Lourenco, M., De Sordi, L. & Debarbieux, L. The diversity of bacterial lifestyles promote bacteriophage production in the gut symbiont Lactobacillus reuteri. Cell hampers bacteriophage tenacity. Viruses 10, E327 (2018). Host Microbe 25, 273–284 e276 (2019). 93. Brussow, H. Bacteriophage-host interaction: from splendid isolation into a messy 63. Samuels, A. N., Roggiani, M., Zhu, J., Goulian, M. & Kohli, R. M. The SOS response reality. Curr. Opin. Microbiol. 16, 500–506 (2013). mediates sustained colonization of the mammalian gut. Infect. Immun. 87, 94. Seed, K. D. et al. Evolutionary consequences of intra-patient phage predation on e00711–e00718 (2019). microbial populations. Elife 3, e03497 (2014). 64. Jubelin G., et al. Modulation of enterohaemorrhagic Escherichia coli survival and 95. Reyes, A., Wu, M., McNulty, N. P., Rohwer, F. L. & Gordon, J. I. Gnotobiotic mouse virulence in the human gastrointestinal tract. Microorganisms 6, pii: E115 (2018). model of phage-bacterial host dynamics in the human gut. Proc. Natl Acad. Sci. 65. Hernandez, S. B., Cota, I., Ducret, A., Aussel, L. & Casadesus, J. Adaptation and USA 110, 20236–20241 (2013). preadaptation of Salmonella enterica to bile. PLoS Genet. 8, e1002459 (2012). 96. Miedzybrodzki, R. et al. Clinical aspects of phage therapy. Adv. Virus Res. 83, 66. Diard, M. et al. Inflammation boosts bacteriophage transfer between Salmonella 73–121 (2012). spp. Science 355, 1211–1215 (2017). 97. Sarker, S. A. & Brussow, H. From bench to bed and back again: phage therapy of 67. Rossmann, F. S. et al. Phage-mediated dispersal of biofilm and distribution of childhood Escherichia coli diarrhea. Ann. N. Y. Acad. Sci. 1372,42–52 (2016). bacterial virulence genes is induced by quorum sensing. PLoS Pathog. 11, 98. Sarker, S. A. et al. Oral phage therapy of acute bacterial diarrhea with two e1004653 (2015). coliphage preparations: a randomized trial in children from Bangladesh. EBio- 68. Silpe, J. E. & Bassler, B. L. Phage-encoded LuxR-type receptors responsive to host- Medicine 4, 124–137 (2016). produced bacterial quorum-sensing autoinducers. MBio 10,e00638–19 (2019). 99. Ott, S. J. et al. Efficacy of sterile fecal filtrate transfer for treating patients with 69. Taylor, V. L., Fitzpatrick, A. D., Islam, Z. & Maxwell, K. L. The diverse impacts of Clostridium difficile infection. Gastroenterology 152, 799–811 e797 (2017). phage morons on bacterial fitness and virulence. Adv. Virus Res. 103,1–31 100. Jahn, M. T. et al. A phage protein aids bacterial symbionts in eukaryote immune (2018). evasion. Cell Host Microbe 26, 542–550 e545 (2019). 70. Wahl, A., Battesti, A. & Ansaldi, M. Prophages in Salmonella enterica: a driving 101. Tetz, G. V. et al. Bacteriophages as potential new mammalian pathogens. Sci. force in reshaping the genome and physiology of their bacterial host? Mol. Rep. 7, 7043 (2017). Microbiol. 111, 303–316 (2019). 102. Galtier, M. et al. Bacteriophages to reduce gut carriage of antibiotic resistant 71. Rolain, J. M., Fancello, L., Desnues, C. & Raoult, D. Bacteriophages as vehicles of uropathogens with low impact on microbiota composition. Environ. Microbiol. the resistome in cystic fibrosis. J. Antimicrob. Chemother. 66, 2444–2447 (2011). 18, 2237–2245 (2016). 72. Fancello, L., Desnues, C., Raoult, D. & Rolain, J. M. Bacteriophages and diffusion 103. Dissanayake, U., Ukhanova, M., Moye, Z. D., Sulakvelidze, A. & Mai, V. Bacter- of genes encoding antimicrobial resistance in cystic fibrosis sputum microbiota. iophages reduce pathogenic Escherichia coli counts in mice without distorting J. Antimicrob. Chemother. 66, 2448–2454 (2011). gut microbiota. Front. Microbiol. 10, 1984 (2019). 73. Debroas D. & Siguret C. Viruses as key reservoirs of antibiotic resistance genes in 104. Febvre, H. P. et al. PHAGE study: effects of supplemental bacteriophage the environment. ISME J. 13, 2856–2867 2019. intake on inflammation and gut microbiota in healthy adults. Nutrients 11, E666 74. Calero-Caceres, W., Ye, M. & Balcazar, J. L. Bacteriophages as environmental (2019). reservoirs of antibiotic resistance. Trends Microbiol. 27, 570–577 (2019). 105. Sarker, S. A. et al. Oral application of Escherichia coli bacteriophage: safety tests 75. Modi, S. R., Lee, H. H., Spina, C. S. & Collins, J. J. Antibiotic treatment expands the in healthy and diarrheal children from Bangladesh. Environ. Microbiol .19, resistance reservoir and ecological network of the phage metagenome. Nature 237–250 (2017). 499, 219–222 (2013). 106. Chan, B. K. et al. Phage selection restores antibiotic sensitivity in MDR Pseudo- 76. Enault, F. et al. Phages rarely encode antibiotic resistance genes: a cautionary monas aeruginosa. Sci. Rep. 6, 26717 (2016). tale for virome analyses. ISME J. 11, 237–247 (2017). 107. Chatterjee, A. et al. Bacteriophage resistance alters antibiotic-mediated intest- 77. Davies, J. & Davies, D. Origins and evolution of antibiotic resistance. Microbiol. inal expansion of Enterococci. Infect. Immun. 87, e00085–19 (2019). Mol. Biol. Rev. 74, 417–433 (2010). 108. Lossouarn, J. et al. Enterococcus faecalis countermeasures defeat a virulent 78. Leigh, B. A. Cooperation among conflict: prophages protect bacteria from picovirinae bacteriophage. Viruses 11, E48 (2019). phagocytosis. Cell Host Microbe 26, 450–452 (2019). 109. Rigottier-Gois, L. et al. The surface rhamnopolysaccharide epa of Enterococcus 79. Feiner, R. et al. A new perspective on lysogeny: prophages as active regulatory faecalis is a key determinant of intestinal colonization. J. Infect. Dis. 211,62–71 switches of bacteria. Nat. Rev. Microbiol. 13, 641–650 (2015). (2015). 80. Rabinovich, L., Sigal, N., Borovok, I., Nir-Paz, R. & Herskovits, A. A. Prophage 110. Kortright, K. E., Chan, B. K., Koff, J. L. & Turner, P. E. Phage therapy: a renewed excision activates Listeria competence genes that promote phagosomal escape approach to combat antibiotic-resistant bacteria. Cell Host Microbe 25, 219–232 and virulence. Cell 150, 792–802 (2012). (2019). 81. Bille, E. et al. A virulence-associated filamentous bacteriophage of Neisseria 111. Golomidova, A., Kulikov, E., Isaeva, A., Manykin, A. & Letarov, A. The diversity of meningitidis increases host-cell colonisation. PLoS Pathog. 13, e1006495 (2017). coliphages and coliforms in horse feces reveals a complex pattern of ecological 82. Obeng, N., Pratama, A. A. & Elsas, J. D. V. The significance of mutualistic phages interactions. Appl Environ. Microbiol. 73, 5975–5981 (2007). for bacterial ecology and evolution. Trends Microbiol. 24, 440–449 (2016). 112. Rodriguez-Valera, F. et al. Explaining microbial population genomics through 83. Duerkop, B. A., Clements, C. V., Rollins, D., Rodrigues, J. L. & Hooper, L. V. A phage predation. Nat. Rev. Microbiol. 7, 828–836 (2009). composite bacteriophage alters colonization by an intestinal commensal bac- 113. Le Chatelier, E. et al. Richness of human gut microbiome correlates with terium. Proc. Natl Acad. Sci. USA 109, 17621–17626 (2012). metabolic markers. Nature 500, 541–546 (2013). 84. Frazao, N., Sousa, A., Lassig, M. & Gordo, I. Horizontal gene transfer overrides 114. Torres-Barcelo, C. The disparate effects of bacteriophages on antibiotic-resistant mutation in Escherichia coli colonizing the mammalian gut. Proc. Natl Acad. Sci. bacteria. Emerg. Microbes Infect. 7, 168 (2018). USA 116, 17906–17915 (2019). 115. Norman, J. M. et al. Disease-specific alterations in the enteric virome in 85. Sekse, C., Solheim, H., Urdahl, A. M. & Wasteson, Y. Is lack of susceptible reci- inflammatory bowel disease. Cell 160, 447–460 (2015). pients in the intestinal environment the limiting factor for transduction of Shiga 116. Majewska, J. et al. Oral application of T4 phage induces weak antibody pro- toxin-encoding phages? J. Appl. Microbiol. 105, 1114–1120 (2008). duction in the gut and in the blood. Viruses 7, 4783–4799 (2015). 86. Zhao, S. et al. Adaptive evolution within gut microbiomes of healthy people. Cell 117. Nguyen S., et al. Bacteriophage transcytosis provides a mechanism to cross Host Microbe 25, 656–667 e658 (2019). epithelial cell layers. MBio 2017 8, e01874–17 (2017). 87. Chibani-Chennoufi, S. et al. In vitro and in vivo bacteriolytic activities of 118. Barfoot, R. et al. Some properties of dendritic macrophages from peripheral Escherichia coli phages: implications for phage therapy. Antimicrob. Agents lymph. Immunology 68, 233–239 (1989). Chemother. 48, 2558–2569 (2004). 119. Wenger, S. L., Turner, J. H. & Petricciani, J. C. The cytogenetic, proliferative and 88. Maura, D. & Debarbieux, L. On the interactions between virulent bacteriophages viability effects of four bacteriophages on human lymphocytes. Vitro 14, and bacteria in the gut. Bacteriophage 2, 229–233 (2012). 543–549 (1978). 89. De Sordi, L., Khanna, V. & Debarbieux, L. The gut microbiota facilitates drifts in 120. Aronow, R., Danon, D., Shahar, A. & Aronson, M. Electron microscopy of in vitro the genetic diversity and infectivity of bacterial viruses. Cell Host Microbe 22, endocytosis of T2 phage by cells from rabbit peritoneal exudate. J. Exp. Med. 801–808 e803 (2017). 120, 943–954 (1964). 90. Gabig, M. et al. The cell surface protein Ag43 facilitates phage infection of 121. Mabbott, N. A., Donaldson, D. S., Ohno, H., Williams, I. R. & Mahajan, A. Microfold Escherichia coli in the presence of bile salts and carbohydrates. Microbiology 148 (M) cells: important immunosurveillance posts in the intestinal epithelium. (Pt 5), 1533–1542 (2002). Mucosal Immunol. 6, 666–677 (2013).

Mucosal Immunology (2020) 13:205 – 215 New insights into intestinal phages R. Sausset et al. 215 122. Yang, J. Y. et al. Enteric viruses ameliorate gut inflammation via Toll-like receptor Open Access This article is licensed under a Creative Commons 3 and Toll-like receptor 7-mediated interferon-beta production. Immunity 44, Attribution 4.0 International License, which permits use, sharing, 889–900 (2016). adaptation, distribution and reproduction in any medium or format, as long as you give 123. Van Belleghem, J. D., Clement, F., Merabishvili, M., Lavigne, R. & Vaneechoutte, appropriate credit to the original author(s) and the source, provide a link to the Creative M. Pro- and anti-inflammatory responses of peripheral blood mononuclear cells Commons license, and indicate if changes were made. The images or other third party induced by Staphylococcus aureus and Pseudomonas aeruginosa phages. Sci. material in this article are included in the article’s Creative Commons license, unless Rep. 7, 8004 (2017). indicated otherwise in a credit line to the material. If material is not included in the 124. Sweere, J. M. et al. Bacteriophage trigger antiviral immunity and prevent article’s Creative Commons license and your intended use is not permitted by statutory clearance of bacterial infection. Science 363, eaat9691 (2019). regulation or exceeds the permitted use, you will need to obtain permission directly 125. Brussow, H. Hurdles for phage therapy to become a reality-an editorial com- from the copyright holder. To view a copy of this license, visit http://creativecommons. ment. Viruses 11, E557 (2019). org/licenses/by/4.0/. 126. Zheng, Q., Chen, Q., Xu, Y., Suttle, C. A. & Jiao, N. A virus infecting marine photoheterotrophic Alphaproteobacteria (Citromicrobium spp.) defines a new lineage of ssDNA viruses. Front. Microbiol. 9, 1418 (2018). © The Author(s) 2020

Mucosal Immunology (2020) 13:205 – 215